As technology continues to rapidly evolve, so do hiring and recruiting practices. A number of start-up companies have emerged in recent years offering employers the ability to use artificial intelligence (AI) to screen job candidates and determine their employability. These AI-driven recruiting practices, such as those that use facial and voice recognition technologies, are touted as a means of lowering recruiting costs and eliminating bias in the hiring process. But there is growing concern that the use of AI may threaten a job candidate’s privacy and might result in the inadvertent perpetuation of discriminatory hiring practices.
These concerns and others were raised in a recent complaint filed with the Federal Trade Commission (FTC), urging an investigation into one such company’s business practices. The complaint was filed by the Electronic Privacy Information Center (EPIC), a public interest watchdog located in Washington, D.C. EPIC’s complaint challenges the AI-driven recruiting solutions developed and sold by a company called HireVue, which currently has more than 700 corporate customers that use its technology as part of their hiring process.
EPIC’s stated mission is to focus public attention on emerging privacy and civil liberties issues. In its thirteen-page complaint, EPIC alleges that HireVue has engaged in unfair and deceptive trade practices in violation of section 5 of the FTC Act, and urges the FTC to initiate an investigation. EPIC further calls on the FTC to issue an injunction restricting further use of HireVue’s technology, and to require HireVue to disclose additional information regarding its technology and confidential algorithms.
Although HireVue denies using facial recognition technology, it does acknowledge that, through online interviews with potential job candidates, its technology collects tens of thousands of data points from each video interview. This information includes a candidate’s intonation, inflection, and emotions. These data points are then plugged into a confidential algorithm, which purportedly gives insight into that applicant’s likelihood of success in the position. Candidates are not given access to their algorithmic assessment scores. According to EPIC’s complaint, HireVue’s customers rely on this AI-driven technology to make decisions about which candidates continue through the recruiting process, and which do not.
EIPC argues in its complaint that the term “facial recognition technology” is not limited to technology used for identification or authentication purposes but also includes technologies that detect and analyze basic human facial geometry and facial biometrics to predict demographic characteristics, expression, or emotions. Under this definition, then, EPIC contends that HireVue’s denial that it uses facial recognition technology is a false statement, and therefore a deceptive trade practice under the FTC Act.
Furthermore, EPIC’s complaint raises serious concerns regarding HireVue’s contention that its technology will help eliminate bias in the hiring process. The accuracy of facial recognition technology has been called into question by multiple studies, which are cited throughout EPIC’s complaint. EPIC alleges that HireVue’s technology and algorithms are not sufficiently tested for reliability, validity, or accuracy, and points to numerous studies that conclude that AI tools often contain implicit gender and racial biases. For example, facial recognition technology analyzes emotion differently based on the individual’s race, reading black faces as angrier than white faces.
EPIC also points out that it is unclear how this technology, which evaluates factors such as a candidate’s eye contact, tone of voice, rate of speech, and body language, would appropriately evaluate individuals with disabilities impacting these data points, such as individuals with autism or Parkinson’s. Individuals with such disabilities may thus be automatically deemed to be less employable based on the way that the AI interprets these factors. Because the algorithms are confidential, job candidates and employers alike are unclear on the exact methodology underlying the technology’s determinations. As such, these determinations cannot be meaningfully challenged.
Furthermore, because the HireVue algorithms are confidential, job candidates are not aware of how their personal data is being used. EPIC contends that HireVue’s “intrusive collection and secret analysis of biometric data…causes substantial privacy harms to job candidates.” Although not specifically addressed in EPIC’s complaint, there has also been growing concern regarding how and for how long personal data is stored, to whom it may be sold, and the potential for data breaches, all of which further compounds the privacy concerns surrounding the use of AI in any context.
It remains to be seen how the FTC will respond to EPIC’s complaint regarding HireVue or if similar complaints against other companies marketing such technology to employers will be forthcoming. In the absence of any federal regulation of facial recognition technology, it can be expected that this area of law will continue to develop rapidly in order to address these concerns.
The first state to specifically address the use of AI in the hiring process is Illinois. Effective January 1, 2020, the Artificial Intelligence Video Interview Act imposes a number of obligations on employers who choose to use AI in their interview process. Employers will be required to notify applicants that AI will be utilized, obtain their consent, and comply with confidentiality and data retention requirements. They will also have to explain how the program works and what characteristics will be evaluated. As the Act currently does not prescribe any penalties for violating it or specify what types of remedies are available, it is unclear how it will be enforced.
Unless and until the federal government provides direction, a patchwork of similar state laws will likely emerge. This will be particularly challenging for employers who operate in multiple states. It is important that all employers who currently use AI as part of their interview process, or intend to do so in the future, become familiar with and adhere to these developing laws in each state in which they operate.